At a first glance, a dataset dominated by Epsilonproteobacteria and Betaproteobacteria, with some CPR phyla (Omnitrophica and Parcubacteria) presence. Sites that smelt like sulphur during sampling are populated by Epsilonproteobacteria, while iron and iron-hydroxide filled sites have strong Betaproteobacterial presence on most occasions, showing Omnitrophica and Parcubacteria occupying close to 30-40% of the taxonomic profiles. Lindsay stands alone, showing CPR phyla occupying most of the profile and low-abundance but numerous phyla representing close to 40% of the profile throughout the year.
But is this seen at lower taxonomic levels?
It seems that things are not that simple in this dataset! To put it simply, Epsilonproteobacteria-rich samples are co-dominated by Sulfuricurvum and Sulfurovum and Betaproteobacteria-rich samples by Gallionella and Sideroxydans. Lindsay, as seen before, is mostly made up of low-abundance taxa (not shown) with CPR phyla representing close to 50% of the profile across the year. Taff’s well, albeit iron-rich, probably due to its external influences being on the edge of the coalfield, is populated by a mix of iron- and sulfur-oxidisers in April, which then turns into a FeOB-dominated profile in August, with Ferriphaselus going up to 50% in December.
It’s unclear why bacteria with very similar metabolisms (iron- and sulfur-oxidisers) would co-occur and seemingly compete for dominance of the profiles this way. What could drive this? What do available genomes for Sulfuricurvum, Sulfurovum, Gallionella, Sideroxydans and Ferriphaselus say?
Gallionella is an microaerophilic iron-oxidiser that uses \(Fe^{+2}\) solely as electron donor. It’s equipped for aerotaxis and motile. Grows preferrably at lower temperatures (6ºC) and is unable to grow at 30ºC. Carbon metabolism works through the RubisCO pathway.
Sideroxydans is a motile microaerophilic iron-oxidiser that uses \(Fe^{+2}\) as electron donor but also has the potential to use thiosulfate. This is due to the presence of a sox operon (ABXYZ) in its genome. Grows preferrably at higher temperatures (30ºC) and is unable to grow at 6ºC. Carbon metabolism works through the RubisCO pathway.
Ferriphaselus is a motile microaerophilic iron-oxidiser that uses \(Fe^{+2}\) solely as electron donor. However, it lacks the Mto gene cluster to oxidise \(Fe^{+2}\) and seems to use instead a panoply of alternative FeO metabolisms specially including act (AB1B2CDEF). Further studies are needed to prove its capacity to grow on sulfur species, since not only it has a sox operon (ABXYZ) in its genome but also dsr (ABEFHCMKJOPN), rdsr (DAGC), Sqr and soe (ABC). Grows preferrably at higher temperatures (30ºC) and is unable to grow at 6ºC. Carbon metabolism works through the RubisCO pathway.
Sulfuricurvum is a motile facultative anaerobe that can use sulfide, sulfite and elemental sulfur as electron donors along with oxygen, nitrate and nitric oxide as electron acceptors. To allow that number of electron donors, Sulfuricurvum not only uses the sox (ABXYZ) operon, but also Sqr (DF), Sor (AB) and Fcc (AB). 4 plasmids were found in Handley et al. (2014) with one associated to nitrogen metabolism. Carbon metabolism works through the rTCA cycle.
Sulfurovum is a non-motile facultative anaerobe that can use thiosulfate and elemental sulfur as electron donors along with oxygen and nitrate as electron acceptors. Sulfurovum only uses the sox (ABXYZ) operon, as opposed to Sulfuricurvum. Carbon metabolism works through the rTCA cycle.
Genome Accessions: NC_014394.1, NC_013959.1, BBTH01000001:BBTH01000023, NZ_CP011308.1, CP003920.1.
Genome annotation done with PROKKA v1.12 (https://github.com/tseemann/prokka), functional annotation with EggNOG-mapper (http://eggnogdb.embl.de/app/emapper#/app/emapper).
| Hits | Gallionella | Sideroxydans | Ferriphaselus | Sulfuricurvum | Sulfurovum |
|---|---|---|---|---|---|
| Mercury | 3 | 0 | 0 | 0 | 0 |
| Arsenic | 3 | 2 | 4 | 3 | 3 |
| Cobalt | 14 | 4 | 6 | 4 | 5 |
| Zinc | 3 | 2 | 2 | 2 | 4 |
| Copper | 1 | 5 | 2 | 3 | 6 |
| Manganese | 1 | 1 | 0 | 1 | 1 |
| Selenium | 1 | 1 | 2 | 0 | 2 |
| Magnesium | 0 | 1 | 2 | 0 | 1 |
| Heavy metal efflux pump/-related | 15 | 5 | 7 | 2 | 3 |
| Mg/Co Transport | 3 | 2 | 2 | 2 | 3 |
| Sodium exchange-related | 6 | 6 | 4 | 1 | 6 |
As advertised, Gallionella seems to be a lot more prepared to thrive in heavy-metal-rich environments. Maybe that gives this genus an advantage in colonizing most of the iron-rich sites in South Wales. Not a lot can be taken from this since these genomes were generated from far away sites across the world. They do, however, reveal the potential in some of these genera for competitive advantages in heavy-metal-rich sites.
Ok then, so do any of the metals we measured have any impact on the relative abundance of any of these genera?
Can we try and get numbers out of these supposed relations?
What follows is a heatmap of Pearson correlations. Pearson assumes linearity between variables as well as normal distributions of data and is sensible to transformation. Normalization of the metadata variables to a 0-1 scale didn’t change results here.
So from Pearson correlations it looks like each of the relative abundance of each FeOB is defined by a specific set of metadata variables that probably drives co-occurrence across the dataset.
Gallionella is very positively correlated to cobalt concentrations and negatively to barium, temperature, pH and sodium concentrations. The first two make sense since this genus is known to grow better a lower temperatures and has had its potential for manganese, cobalt and arsenic resistance reported here.
Regarding Sideroxydans, it seems the only meaningful correlations point positively to cobalt and copper negatively to pH and \(Hg^{200}\). Again, at least some part of this may have been predicted by previous work. It’s positively correlated to \(SO^{4}\), which might indicate the presence of a sox operon as in the genome analysed previously.
Ferriphaselus stands out as being positively correlated to cobalt, like Gallionella, and negatively correlated with Na and Mg.
A number of correlations arise with Sulfuricurvum, where it seems that it’s strongly associated to dissolved oxygen, boron, lithium and mercury concentrations. It’s negatively associated to the presence of copper and selenium.
Sulfurovum presence is by positive correlations towards temperature, pH, B, Mg, and \(Hg^{200,201}\). It’s also strongly negatively correlated to copper and manganese presence.
Even though of course for most of it not as many correlations come up, some interesting ones still do. Namely, most of the CPR phyla show some positive correlation towards arsenic. No general distinctions come up regarding phylum-level taxonomic differences. Temperature doesn’t seem to be related at all with CPR relative abundance.
Interestingly, Cand. Falkowbacteria has the strongest positive correlation with barium, Cand. Moranbacteria with zinc, Cand. Azambacteria with lithium and Omnitrophica Inc. Sedis with selenium.
Any ´evidence´ for this in the genomes?
| Hits | Omnitrophica | Azambacteria | Jorgensenbacteria | Nomurabacteria | Moranbacteria | Magasanikbacteria | Wolfebacteria |
|---|---|---|---|---|---|---|---|
| Mercury | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| Arsenic | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| Cobalt | 7 | 5 | 0 | 0 | 0 | 0 | 0 |
| Zinc | 2 | 0 | 3 | 1 | 2 | 3 | 3 |
| Copper | 1 | 0 | 0 | 1 | 2 | 0 | 1 |
| Manganese | 2 | 0 | 0 | 1 | 1 | 1 | 0 |
| Selenium | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| Magnesium | 1 | 1 | 0 | 1 | 1 | 1 | 0 |
| Heavy metal efflux pump/-related | 2 | 0 | 0 | 2 | 2 | 1 | 1 |
| Mg/Co Transport | 1 | 1 | 0 | 1 | 1 | 1 | 0 |
| Sodium exchange-related | 1 | 0 | 2 | 1 | 0 | 2 | 1 |
| Chemotaxis | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| Pilli | 2 | 3 | 4 | 9 | 8 | 9 | 10 |
| Flagella | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
CPR phyla are supposed to mostly be intracellular parasites. Do their relative abundances correlate to those of the main genera?
# cocor_gen = as(metadata_cor_main_genera[, c("Genus", "Abundance")], "data.frame")
# names(cocor_gen) = c("Taxa","Abundance")
# # cocor_gen$type = "CPR"
# cocor_cpr = as(metadata_cor_cpr[, c("Class", "Abundance")], "data.frame")
# names(cocor_cpr) = c("Taxa","Abundance")
# cocor_cpr$type = "Main Genera"
#
# ldply(list(cocor_gen = cocor_gen, cocor_cpr = cocor_cpr))
# cocor_df = rbind(cocor_gen,cocor_cpr)